development of integrated framework for distributed multi...

6th International Conference on Advances in Experimental Structural Engineering

11th International Workshop on Advanced Smart Materials and Smart Structures Technology

August 1-2, 2015, University of Illinois, Urbana-Champaign, United States

Development of Integrated Framework for

Distributed Multi-Platform Simulation

X. Huang1, V. Sadeghian2, O. Kwon3 1 PhD Candidate, Department of Civil Engineering, University of Toronto, Toronto, Canada

Email: [email protected]

2 PhD Candidate, Department of Civil Engineering, University of Toronto, Toronto, Canada


3 Assistant Professor, Department of Civil Engineering, University of Toronto, Toronto, Canada


ABSTRACT

A large-scale numerical/experimental distributed simulation framework is being developed at the University of

Toronto to facilitate the simulation of coupled systems including diverse substructure modules and integration

modules. Unlike the former frameworks for hybrid simulation which focuses on a dedicated integration module

for hybrid simulation, the proposed development focuses on the definition of flexible data exchange formats and

a communication protocol through which any potential integration modules and substructure modules will be

interoperable. Communication between an integration module and substructure modules is established through an

interface element in which the standardized data exchange format and communication protocol are implemented.

The data exchange format is designed in such a general way that it can flexibly transfer all information needed for

various types of simulations such as real-time simulation, hybrid simulation with thermal load, etc. The

development of the interface element and the data exchange format is illustrated through an OpenSees-based

integration module, one of the integration modules implemented in the proposed simulation framework. Retaining

the characteristics of OpenSees, the integration module allows either serial or parallel computing algorithms to be

used for distributed simulation. To illustrate the current developments, a simulation example with soil-structure

interaction analysis using the proposed simulation framework is presented in this paper.

KEYWORDS: Multi-Platform Simulation, Distributed Simulation, Substructure, Parallel Computing

1. INTRODUCTION

The hybrid (experiment-analysis) simulation method has been extensively investigated over the past decades. The

focus of the investigations has been on the development of robust time integration schemes (Chang et al. 2011;

Chen and Ricles 2008; Hung and El-Tawil 2009), control methods to compensate the actuator delay in real-time

hybrid simulations (Carrion and Spencer, Jr. 2007; Darby et al. 2002; Horiuchi et al. 1999; Mercan and Ricles

2009; Phillips and Spencer 2012), and the development of framework for the integration of diverse substructure

modules (Karavasilis et al. 2008; Kwon et al. 2008; Saouma et al. 2012; Schellenberg et al. 2009). This research

has contributed to the application of the hybrid simulation method in practical research projects.

In the hybrid simulation, the governing equation of motion (Eq. 1.1) is solved numerically using the time stepping

method. In the numerical time stepping method, the restoring force term, ( , )r u u , is obtained either from physical

specimens or substructure modules. The restoring force term characterizes the unique inelastic cyclic response of

structural elements which influences the dynamic response of the structural system. Thus, to accurately capture

the dynamic structural response, it is important to realistically evaluate the restoring force term either through

physical specimens or realistic nonlinear structural elements.

( , ) mu cu r u u f (1.1)

Several simulation frameworks have been developed to carry out hybrid simulations. As illustrated in Figure 1.1,

key components of the hybrid simulation frameworks are: 1) a numerical time integration scheme, 2) numerical

elements that can model inelastic cyclic behaviour of the rest of the system, 3) a method to interface with physical

specimens, 4) an actuator controller and a physical specimen, and 5) a method for delay compensation in the case

of real-time hybrid simulations. Depending on the application of the framework, these components are

implemented in various ways. For example, in the hybrid simulation framework, UI-SimCor (Kwon et al. 2008),

an explicit time integration scheme (α-OS method, Combescure and Pegon 1997) is used as a main time-

integration scheme. The restoring forces from physical specimens or substructure modules are obtained through

the computer network. The experimental facility is expected to have an ability to generate ramp and execute the

displacement command from UI-SimCor. The framework has been applied to many hybrid simulations because

it is open source, and also allows integration of diverse substructure elements. Yet, because all restoring forces

are obtained from substructure modules through the computer network, the framework suffers from an increase

in network traffic and computing time when a large number of degrees of freedoms are used in the simulation.

OpenSees (Mazzoni et al. 2007) is an open source seismic analysis tool with many unique inelastic structural

elements. The software also has been used in many hybrid simulations. OpenSees is a self-contained analysis tool

which includes time integration schemes and numerical elements. A middleware which is termed as OpenFresco

(Schellenberg et al. 2009) is used to communicate with actuator controllers. Other frameworks, such as Hybrid

FEM (Karavasilis et al. 2008) or Mercury (Saouma et al. 2012) implement nonlinear structural elements, and are

aimed at running real-time simulations.

Figure 1.1 Components of the hybrid simulation framework

The basis of the hybrid simulation method is very similar to the substructure analysis method in which a large

structural system is partitioned into multiple smaller substructures, each of which is analyzed separately. The

substructure analysis method has been mainly used to partition a large structural system with smaller substructures

such that distributed computational resources can be used. A major difference between the hybrid (experiment-

analysis) simulation and the substructure analysis method is that at least one substructure is represented as a

physical specimen. In addition, unlike purely numerical substructure analyses, the hybrid simulation does not

allow iterations within a time step. This feature, however, is beneficial when integrating numerical models from

different analysis tools. Because of legacy or intellectual property issues, it is very difficult to rewrite and integrate

models from multiple analysis tools, such as Abaqus (Hibbit et al. 2001), OpenSees (Mazzoni et al. 2007), VecTor

(Wong and Vecchio 2002), or Ansys. If the hybrid simulation method is extended for numerical substructure

simulation, a complex structural system, such as the soil-foundation-structure system or a structure with shear-

critical reinforced concrete elements, can be analyzed more accurately by modelling each component with a more

appropriate tool for the component and by integrating them.

The key requirement for such integration is a standardized data exchange format and a communication protocol

through which any numerical analysis tool can integrate other numerical or physical models. This will allow for

maximizing the use of available analysis tools and the use of computational and experimental resources. The main

objective of the research presented in this paper is to develop a standardized data exchange format which can

greatly facilitate integration of diverse numerical analysis tools and physical specimens. In Section 2, the overall

architecture of the integrated simulation is presented. The architecture does not propose (or oppose) the use of one

analysis tool for the integrated simulation. Rather, it shows how diverse tools can be integrated by using the

proposed data exchange format and its implementation. In Section 3, two analysis examples are presented,

followed by the conclusions in Section 4.

2. INTEGRATED SIMULATION METHOD

2.1 Overall Architecture

At any given time step, the numerical integration scheme (Component 1 in Figure 1.1), assembles a global force

vector from the numerical elements, and solves for the unknown state variables (i.e., displacement, velocity, and

Numerical substructure elements

Physically

represented

Middleware for controller

Numerical

model

Physical substructure element

PID loop

Numerical integration

scheme

Component #1 Component #2

Component #3Component #4

Controller

acceleration vectors) in the time step. Depending on the nonlinear solution scheme, it might be necessary to

assemble a global tangent stiffness matrix. The numerical elements may require predicted displacements and/or

velocities to return force vectors and stiffness matrices. By collecting the force vectors and stiffness matrices from

different numerical elements or analysis tools at each time step, it is possible to find unknown state variables and

move on to the next time step. A main functionality that is required for this integration is the communication

between numerical models (or analysis tools) at each time step and the numerical integration scheme.

In this study, a few assumptions are made to integrate diverse structural analysis tools. First, it is assumed that all

substructure modules are either displacement or velocity dependent, and they do not have their own mass. This

assumption simplifies the integration of various components; if the substructure represents a mass-damper-spring

system, then each substructure module needs to run dynamic analyses which can introduce stability issues in the

dynamic integration scheme. This issue was observed in the dynamic soil-structure-interaction analysis in which

the soil-foundation system is modelled based on dynamic impedance functions (Lesgidis et al. 2015). Based on

the assumption, one can classify the analysis tools with two components: 1) an integration module which runs

time integration schemes with the numerical model of majority of structural systems (Component #1 and #2 in

Figure 1.1), and 2) static substructure modules which run static analysis or experiment (Component #3 and #4 in

Figure 1.1). The second assumption is that the nonlinear solution scheme in the integration module does not take

a trial and error approach. Depending on the nonlinear solution scheme, an incremental displacement is tried and

cancelled later on, if necessary, to try a smaller incremental displacement. This method is frequently used when

the model cannot converge to a solution in a time step. This method, however, is not applicable to experimental

elements or to a substructure element modelled in a different analysis tool. Thus, at this stage, it is assumed that

the integration module runs a non-iterative explicit time integration scheme.

Figure 2.1 Integration and substructure modules

With the two assumptions in the above, any large structural system can be analyzed by integrating more than one

analysis tool, including an integration module and substructure modules. Once one of the substructure modules

are replaced with a physical specimen, then it becomes a hybrid (experiment-analysis) simulation. Otherwise, the

analysis becomes a multi-platform simulation. Several integration and substructure modules can be integrated

together using the above framework. The current development is illustrated in Figure 2.1. The authors

implemented standardized communication functionality in the following integration modules: UI-SimCor (Kwon

et al. 2008), Cyrus (Sadeghian et al. 2015a; b), and OpenSees and OpenSeesSP (Mazzoni et al. 2007). Cyrus has

been developed as a main integration module for VecTor programs which specialize in analyzing reinforced

concrete structures based on Modified Compression Field Theory and Disturbed Stress Model (Vecchio and

Collins 1986; Vecchio 2000). In addition to integrating the VecTor programs, a research is in progress to use

Cyrus as a main integration module for other substructure modules including physical specimens. If the majority

of the structural system is made of reinforced concrete, then Cyrus is expected to be most reliable analysis tool.

To use OpenSees and OpenSeesSP as a main integration module, a SubStructure element in the program is

modified such that it can collect the substructure model’s restoring force and stiffness matrix through the network.

In addition, the port tunnelling method is used such that OpenSeesSP operating in a SciNet supercomputer can

OpenSeesCluster or Dekstop

Subs.Elem.

Zeus-NL

Abaqus

VecTor 2, 3, 4, and 5

Actuators, MAST, UT10 Simulator

OpenSees

UI-SimCor v3.0

DesktopNICON

NICA

NICA

Integration Modules

Numerical Substructure Modules

Voltage

Named pipes

UTNP via TCP-IP

Cyrus

NICA

Experimental Substructure Modules

Communication Methods

S-Frame(in progress)

Abaqus(in progress)

Soil-Foundation System(in progress)

communicate with a substructure module in a desktop computer or with a physical specimen. This feature will

allow simulation of a large soil-structure system with most reliable analysis tools and large computing power. In

collaboration with an industry partner, a research project is in progress to use the S-Frame commercial

design/analysis software as an integration module, which will facilitate the use of advanced analysis methods in

the design process.

Several substructure modules can be integrated with the integration modules. For now, the following analysis

tools can be used as substructure modules: OpenSees, VecTor programs, Zeus-NL (Elnashai et al. 2002), Abaqus,

and experimental elements. It is worth noting that OpenSees and Abaqus can be used both as integration and

substructure modules. To interface an experimental element, a unique middleware program, NICON, is developed

which can serve as an interface for up to six coupled DOFs or for up to ten uncoupled DOFs (Zhan and Kwon

2015).

The main requirement for the interoperability between diverse substructure and integration modules is a unified

data exchange format and communication protocol. In addition, each substructure module should have a

functionality to impose the predicted displacement to the numerical or physical model. Each integration module

should have a functionality to collect a force vector or a stiffness matrix from the substructure module as well. In

the next subsection, a standardized data exchange format and its implementation are introduced.

2.2 Standardized Data Exchange Format

To efficiently integrate diverse substructure modules, it is essential to use a standardized data exchange format.

While there have been extensive research projects on hybrid simulation methods, the simulation frameworks have

their own communication method or data exchange format. There were attempts to develop a standardized

communication protocol, NEESgrid Teleoperations Control Protocol (NTCP, Pearlman et al. 2004) and NEES

Hybrid Communications Protocol (NHCP, Cowart et al. 2007) with the support of the NEES program. NHCP is

a successor of NTCP, intended to overcome the limitations of the NTCP. Up to the extent of the authors’

knowledge, however, implementation and application of the NHCP protocol in hybrid simulations is limited.

NHCP includes many useful functionalities which can ensure secure transmission of data packets.

To allow communication between two modules, it is necessary to define the data exchange format and the

communication protocol. Because all data communicated through the network is serialized, both sides should

understand how arrays or vectors are serialized, and what type of data is included in the packet. In addition, both

sides also need to agree on the sequence of data exchange because when one side sends data, the other side should

receive and process the data. In the current implementation, it is assumed that the integration module sends a

command and the substructure module responds to the command. Thus, the protocol is rather straight forward.

This protocol can be improved in future implementations by allowing multiple channels of communication, or by

using UDP instead of TCP-IP to allow faster (yet somewhat unreliable) communication. Further development of

communication protocols is left for a later study. The following discussion is mainly concerned with the

standardized data exchange format.

The University of Toronto Networking Protocol (UTNP) data exchange format consists of a header and

corresponding data. All data is exchanged in binary to reduce the packet size, not to lose accuracy during the

binary-ASCII conversion process, nor to waste processing time during the conversion process. The header consists

of 16 total bytes of data as illustrated in Figure 2.2. Specifically, the version parameter indicates the version of

the data exchange format for the purpose of maintenance. The command parameter indicates the communication

action which can be sending target displacement to the substructures or asking them for restoring forces. The test

type parameter indicates whether the simulation is pseudo-dynamic or real-time. The substructure type describes

the type of the substructure used in the simulation which can be either numerical or experimental. The precision

parameter defines the precision of data appended to the message header. The data type indicates the type of the

appended data, either displacement, force, velocity, acceleration, temperature, or any combination thereof. The

size of the header is fixed to be 16 bytes while the size of the corresponding data depends on the parameters

defined in the message header. For example, if an experimental two-dimensional truss element is represented, the

Number of DOF parameters is 4 (the total number of DOFs of the substructure), the command parameter is 3

(sending target displacements to the substructure), and the precision parameter is 2 (double precision, 8 bytes).

The size of the data appended to the message header will be set to be 32 bytes (4 DOFs × 8 bytes) for target

displacements. More details on the data exchange format will be released to the research community in upcoming

publications.

Figure 2.2 The data exchange format in UTNP

To facilitate the implementation of the above data exchange format, the authors developed a set of functions which

can be used to send and receive data packets using the above data exchange format. A dynamic link library (DLL)

is compiled with the functions, which has been tested with C++ for OpenSees and OpenSeesSP, FORTRAN for

VecTor2, MATLAB for UI-SimCor, and LabView for NICON. Most of the data exchange in Figure 2.1 is carried

out using the data exchange format. Once the robustness of the data exchange function is fully tested, the library

will be open to the research community such that any other research groups can develop their own integration or

substructure modules, which can work together with all the modules illustrated in Figure 2.1.

3. VERIFICATION EXAMPLES

3.1 A Frame with Shear- and Flexure-critical Columns (OpenSees and VecTor2)

Two verification examples are presented in this paper to illustrate current developments. The first example is a

shaking table test of a three bay frame tested by Wu et al. (2009). The specimen consists of two shear-critical and

two flexure-critical columns. The columns are fixed at the bottom to supports at the top to a stiff beam. All four

columns have the same aspect ratio but different reinforcement details are used to obtain both shear and flexural

failure modes. A total of 22.7 tonnes of weight, including the weight of the beam, are placed to simulate gravity

load and inertial mass. As shown in Figure 3.1, one shear-critical column of the structure is modelled as a

substructure with VecTor2 while the rest of the structure is modelled with OpenSees acting as an integration

module. For the OpenSees model, all masses are lumped at the beam-column joints, which are assumed to be

rigid. Each component is represented with one force-based beam column element with five integration points.

Rayleigh damping is used with a damping ratio of 1.8% at the natural period of the structure. For the VecTor2

model, concrete and longitudinal reinforcement is represented by rectangular continuum elements and discrete

truss elements, respectively. Out-of-plane and in-plane transverse reinforcements are smeared over the concrete

core. To accurately simulate the interface between frame elements and continuum elements, frame elements with

three DOFs per node and continuum elements with two DOFs per node are connected through rigid beam elements

as shown in Figure 3.1. Figure 3.2 compares time histories of inter-story drifts measured from experiments and

results from the numerical model. From this figure it can be found that the numerical model underestimates the

displacement response. One potential reason for this would be modelling of the top beam as rigid which may not

properly re-distribute the axial loads to columns after failure. Figure 3.3 shows the deformed shape of the structure

and crack pattern of the shear-critical column modelled with VecTor2 at the maximum lateral displacement

Figure 3.1 Numerical models of the tested frame

Version Command Test type Substructure type

Precision Data type Number of DOFs

Step number Reserved for future extension

Time stamp

Data

32 bit

16 bytes

Size depends

on header

Truss elements

for rebars

Core concrete

Cover concreteRigid elements

Frame elementsLumped masses

OpenSees model VecTor2 model

Figure 3.2 Inter-story drift ratios from numerical models and test specimens

Figure 3.3 Deformed shape and crack pattern at maximum lateral displacement

3.2 Underground Structure Analysis (OpenSees and VecTor2)

The second example is a soil-structure interaction analysis of a subway structure. This verification example is not

aimed to duplicate the actual response of the structure but illustrate the potential applicability of the proposed

simulation framework to future soil-structure interaction analyses of underground structures. A 1/30 scaled

subway station structure tested on a shake table in a centrifuge is selected as the reference structure (Guoxing et

al. 2015). For the purpose of illustration, only one-story one-bay of the structure is considered and is modelled

with VecTor2. Structure dimensions and reinforcement details of the centrifuge model can be found in Figure 3.4.

The soil domain is assumed to be linear elastic and is modelled with OpenSees using a four-node quadrilateral

element as shown in Figure 3.5. Material properties of soil and structure are listed in Table 3.1. There are 86 nodes

with a total of 172 DOFs at the interface between OpenSees and VecTor2 models. Due to the thinness of the RC

structure, structural masses are lumped at the interface nodes in the OpenSees model as shown in Figure 3.5. Taft

ground motion recorded at Taft Lincoln School during the Kern Country earthquake is used in this example. The

ground motion is scaled with peak acceleration of 0.3g and it is applied along the base of the soil domain. The

nodes at the left and right boundaries of the soil domain are tied together in both horizontal and vertical directions.

Rayleigh damping of 1.9% at the first and second mode is assumed. The predicted deformed shape and the crack

pattern are shown in Figure 3.6. It can be found that shear cracks with a maximum crack width of 1.2 mm develop

in the middle span of the bottom slab. The analysis method will be used to investigate the effects of the soil-

structure of underground structures with realistic soil models in OpenSees and reinforced concrete structures in

VecTor programs.

Figure 3.4 Dimensions and reinforcement details of the subway structure

-6

-4

-2

0

2

4

6

0 2 4 6 8 10

Dri

ft, %

Time, sec

Experimental

Numerical


287 mm

25

1 m

m

23

mm

26 mm

∅0.9@50

∅0.9@25

∅0

.9@

12

∅0.9@50

∅0.9@25

Table 3.1 Material properties of model soil and structure (1/30 scale model)

Soil

Elastic Modulus (MPa) 8.40

Structure

Elastic Modulus of Concrete (MPa) 7500

Density (ton/m3) 1.76 Concrete Compressive Strength (MPa) 7.5

Shear Wave Velocity (m/s) 40 Elastic Modulus of Rebars (MPa) 50000

Poisson’s Ratio 0.49 Yielding Strength of Rebars (MPa) 100

Figure 3.5 Numerical models of subway station structure

Figure 3.6 Deformed shape and crack pattern of the structure

4. CONCLUSION

This paper presents a research project in progress at the University of Toronto. As a first step toward developing

an integrated simulation framework, the authors defined a standardized data exchange format which was

implemented in various analysis tools. In addition, a method was developed to allow integration of a model in a

computer cluster (OpenSeesSP) with a model in a desktop computer. The interoperability will allow for

maximized use of various analysis tools and computational/experimental resources. Two analysis examples are

presented in this paper clearly demonstrating the potential of the integrated simulation method. Further research

is in progress to allow for the integration of dynamic substructure modules.

ACKNOLWEDGEMENT The research project was funded by Ontario Early Researcher Award and Discovery Grant from Natural Sciences

and Engineering Research Council of Canada.

REFERENCES

1. Carrion, J. E., and Spencer, Jr., B. F. (2007). Model-based Strategies for Real-time Hybrid Testing. Urbana.

2. Chang, S.-Y., Yang, Y.-S., and Hsu, C.-W. (2011). “A family of explicit algorithms for general pseudodynamic

testing.” Earthquake Engineering and Engineering Vibration, 10(1), 51–64.


Continuum element

Quad element

Interface nodes

14

0 m

m2

51

mm

40

0 m

m

547 mm 287mm 547 mm


3. Chen, C., and Ricles, J. M. (2008). “Development of Direct Integration Algorithms for Structural Dynamics

Using Discrete Control Theory.” Journal of Engineering Mechanics, 134(8), 676.

4. Combescure, D., and Pegon, P. (1997). “α-Operator splitting time integration technique for pseudodynamic

testing error propagation analysis.” Soil Dynamics and Earthquake Engineering, 16(7-8), 427–443.

5. Cowart, C., Hubbard, P., Miller, L., and Crawford, G. (2007). NHCP Reference Implementation and Protocol

Reference Guide.

6. Darby, a. P., Williams, M. S., and Blakeborough, A. (2002). “Stability and Delay Compensation for Real-Time

Substructure Testing.” Journal of Engineering Mechanics, 128(12), 1276.

7. Elnashai, A. S., Papanikolaou, V. K., and Lee, D. H. (2002). Zeus NL- A System for Inelastic Analysis of

Structures.

8. Guoxing, C., Su, C., Xi, Z., Xiuli, D., Chengzhi, Q., and Zhihua, W. (2015). “Shaking-table tests and numerical

simulations on a subway structure in soft soil.” Soil Dynamics and Earthquake Engineering, Elsevier, 76, 13–

28.

9. Hibbit, Karlsson, and Sorensen. (2001). ABAQUS theory manual. Version 6.2.

10. Horiuchi, T., Inoue, M., Konno, T., and Namita, Y. (1999). “Real-time hybrid experimental system with

actuator delay compensation and its application to a piping system with energy absorber.” Earthquake

Engineering & Structural Dynamics, 28(10), 1121–1141.

11. Hung, C.-C., and El-Tawil, S. (2009). “Full operator algorithm for hybrid simulation.” Earthquake

Engineering & Structural Dynamics, 38(13), 1545–1561.

12. Karavasilis, T. L., Seo, C.-Y., and Ricles, J. (2008). HybridFEM: A Program for Dynamic Time History

Analysis of 2D Inelastic Framed Structures and Real-Time Hybrid Simulation. Bethlehem, PA.

13. Kwon, O. S., Elnashai, A. S., and Spencer, B. F. (2008). “A framework for distributed analytical and hybrid

simulations.” Structural Engineering and Mechanics, 30(3), 331–350.

14. Lesgidis, N., Kwon, O., and Sextos, A. (2015). “A time-domain seismic SSI analysis method for inelastic

bridge structures through the use of a frequency-dependent lumped parameter model.” Earthquake Engineering

and Structural Dynamics, (Accepted).

15. Mazzoni, S., McKenna, F., Scott, M. H., and Fenves, G. L. (2007). “The OpenSees command language manual,

version 2.0.” Pacific Earthquake Engineering Research Center, University of California at Berkeley, Berkeley,

CA.

16. Mercan, O., and Ricles, J. M. (2009). “Experimental Studies on Real-Time Testing of Structures with

Elastomeric Dampers.” Journal of Structural Engineering, 135(9), 1124.

17. Pearlman, L., D’Arcy, M., Johnson, E., Kesselman, C., and Plaszczak, P. (2004). “NEESgrid Teleoperation

Control Protocol (NTCP).”

18. Phillips, B., and Spencer, B. (2012). “Model-Based Feedforward-Feedback Actuator Control for Real-Time

Hybrid Simulation.” Journal of Structural Engineering.

19. Sadeghian, V., Vecchio, F., and Kwon, O. (2015a). “A Framework for Multi-Scale Analysis of Reinforced

Concrete Structures.” ASCE Journal of Structural Engineering, (in review).

20. Sadeghian, V., Vecchio, F., and Kwon, O. (2015b). “An Integrated Framework for Analysis of Mixed-Type

Reinforced Concrete Structures.” CompDyn 2015, Crete, Greece, 13.

21. Saouma, V., Kang, D.-H., and Haussmann, G. (2012). “A computational finite-element program for hybrid

simulation.” Earthquake Engineering & Structural Dynamics, 41(3), 375–389.

22. Schellenberg, A. H., Mahin, S. A., and Fenves, G. L. (2009). Advanced Implementation of Hybrid Simulation.

23. Vecchio, F. J. (2000). “Disturbed Stress Field Model for Reinforced Concrete: Formulation.” Journal of

Structural Engineering, 126(9), 1070–1077.

24. Vecchio, F. J., and Collins, M. P. (1986). “The Modified Compression-Field Theory for Reinforced Concrete

Elements Subjected to Shear.” ACI Journal, 83(2), 219–231.

25. Wong, P. S., and Vecchio, F. J. (2002). VecTor2 & FormWorks User’s Manual. Toronto.

26. Wu, C., Kuo, W., Yang, Y., Hwang, S., Elwood, K. J., Loh, C., and Moehle, J. P. (2009). “Collapse of a

nonductile concrete frame : Shaking table tests.” Earthquake Engineering and Structural Dynamics, 38, 205–

224.

27. Zhan, H., and Kwon, O.-S. (2015). “Actuator controller interface program for pseudodynamic hybrid

simulation.” 2015 World Congress on Advances in Structural Engineering and Mechanics, Songdo, Korea,

(paper submitted).

development of integrated framework for distributed multi...

Documents